metadata
language:
- en
configs:
- config_name: default
data_files:
- split: train
path: newsqa_cleaned_train.parquet
- split: validation
path: newsqa_cleaned_validation.parquet
- split: test
path: newsqa_cleaned_test.parquet
size_categories:
- 10K<n<100K
Converting newsqa dataset to identical format as lmqg/qag_squad for asahi417/lm-question-generation
Questions, that dont end on "?" have been removed.